System requirements

CrestBid performs AI inference locally, so performance depends mostly on your GPU’s VRAM or your Mac’s unified memory — not on your system RAM. It runs on Windows, macOS, and Linux.

We recommend running a 27B LLM, which runs comfortably on a machine with 48 GB of VRAM. Our products are also designed to work with smaller models, with some compromise on speed and accuracy.

Requirements

Operating system	Windows 11 (installer sets up WSL2 and Docker for you) or a modern Linux distribution. Macs run in cloud-model mode — see below.
Disk	60 GB free for the application and its models; more for large knowledge bases.
Network	Only needed for the initial download and setup. After that, CrestBid runs with no internet connection at all.

An NVMe SSD is recommended for fast knowledge-base indexing. CPU-only operation is supported but is much slower than running on a GPU.

No NVIDIA GPU? (including Macs)

Machines without an NVIDIA GPU — including all Macs today — run CrestBid in cloud-model mode: the platform installs and works the same, but inference is served by a model provider you configure, or by a GPU host elsewhere on your own network (remote model mode). It works well for trying the product out; for fully local, private AI, use a Windows or Linux PC with a qualifying NVIDIA GPU. Native Apple-silicon (Metal) inference, using a Mac’s unified memory directly, is on our roadmap.

When a cloud model provider is configured, prompts leave your machine for that provider. For zero data exposure, run local models on a qualifying NVIDIA GPU — or point CrestBid at a GPU host inside your own network.